Stator winding/core failure

Kent Smith, Duke Energy and chairman, GUG steering committee

Smith’s presentation focused on an April 2014 incident in which a 790-MVA generator tripped on ground relay. This failure occurred after 44 years of service. The generator had experienced a 5-min offline over-flux incident in March 1998, ranging between 8% and 25% over-flux. In May 2005, seven instantaneous over-flux events occurred, ranging from 8% to 15% over-flux. The in-service failure occurred nine years after the last over-flux incident. Note that a hiatus between over-flux and failure had been experienced previously in the industry.

None of the over-flux events at this plant was severe, although at 20% over-flux, core iron saturation would be significant.

The winding failed near the end of the core in slot 18 (Fig 1); some core grinding had been done before this photo was taken. The extent of heat penetration at the core melting locations is evident in Fig 2. Finally, there was burn evidence at many locations on the key bars, greatest at the core ends (Figs 3, 4).

To make a reliable repair, the core was removed, restacked with new laminations, and a new winding installed—a long, complicated process. It was necessary to build a temporary structure for workspace. This was a “cage” machine and the inner frame and core required removal, a difficult task on such a large and heavy component. Special equipment was required for rotation up and down. The temporary structure with the up-ended inner frame and new core is shown in Fig 5.

After return to horizontal and reinsertion of the inner frame/core component, a new winding was installed. The outage was about a year in length, but the stator essentially was returned to as-new condition. Access presentation.

Core mechanical isolation problems

Marques Montes, Arizona Public Service Co

Redhawk Generating Station, equipped with two F-class 2 × 1 combined cycles, entered commercial service early in 2003. From the beginning, the generators serving the two steam turbines (STs) were exceptionally noisy—about 116 dBA at all load conditions.

In June 2003, the OEM recommended immediate shutdown for stator end-winding modifications following the failure of another unit in the fleet because of end-winding vibration. With summer peak-load demand a concern, APS proposed instrumenting the end windings with vibration detectors, monitoring the values, and shutting down if the safe limit was exceeded. This was acceptable to the OEM.

The monitor readings were marginally acceptable. In April 2009, the ST generator for Power Block 2 went single phase to ground because of a stator-bar fracture. The winding tripped immediately on ground relay, but current continued to flow through the arc as field current decayed. Damage was substantial (Fig 6).

A third-party service company repaired the winding by splicing in new copper strands and insulating the repaired bar area with mica tape. Additional support components were added and the winding continues to operate safely.

In March 2013, a cooler leak developed following a vane fracture from high vibration. While performing this repair, the cooler was removed, allowing access to the back of the core. Inspection of the structure revealed there was no core vibration isolation horizontally to the frame (Fig 7). Specifically, the core was supported by key bars, the key bars were connected directly to a section plant, and the section plates were connected directly to the frame. The frame connection points are at the six axial locations seen in Fig 8.

Options were evaluated—including harmonic dampeners and wrapper reinforcements proposed by the OEM. None of these options seemed practical or effective. New generators were purchased from another OEM and these replacements are operating quietly. Access presentation.

GVPI issues

Leopoldo Duque Balderas, Comego SA de CV (Mexico) and member, GUG steering committee

Comego recently has had considerable difficulties with a line of globally vacuum pressure impregnated (GVPI) 187-MW, 16.5-kV generators after about 13 years of reliable service. Problems included in-service failures, hipot failures, and major inspection concerns. No satisfactory corrective action short of difficult stator rewind or replacement was available.

Balderas described the GVPI process in detail, offering drawings of the various steps, and discussing the pros (low cost, mechanically sound) and cons (danger of voids, which may lead to severe PD attack) of the GVPI stator winding.

In the process of failure root-cause investigation, severe internal and external PD damage was found on high-voltage (HV) bars. External damage to the surface tapes and grounding paint is clearly visible in Fig 9. The total destruction of internal insulating materials can be seen in Fig 10. Note: Cross-over putty is completely eaten away and the strand insulation is heavily attacked. In the Fig 11 cross section, voids at the bottom of the bare bar caused by internal PD are visible.

Corrective action still was being considered at the time of the meeting. Reversing of high- and low-voltage leads is no practical because of the extent of damage suffered by the windings. The procurement cycle for replacement stators is long and it is doubtful the existing windings can be repaired such as to remain in service until new stators can be obtained. In sum, this line of GVPI generators represents a major maintenance challenge to users.

Stator magnetic cores: design, deterioration, failure

Clyde V Maughan, Maughan Generator Consultants

This presentation summarized core design and duties, emphasizing the sometimes less-well-understood aspects. Maughan began by noting that the function of core iron to shield copper strands from electromagnetic forces often is not discussed. But except for this feature, no insulation system in use today could survive.

As shown in Fig 12, the magnetic flux moves quickly into the tooth iron after crossing the air gap. On hydro-generators, with the wedge at the very top of the tooth, some small amount of flux may intersect the upper-most strands and cause the strand displacement illustrated in Fig 13. With this in mind, there is a concern relative to the top strands of the top bars at the core ends in generators of recent design with large core-iron stepdown at the core ends (Fig 14).

Maughan’s presentation focused primarily on the more recent, and not yet well understood, problem of core burning and failure attributed to key-bar currents. OEMs have been using designs for over 15 years with reduced depth of core back-iron. The result is increased leakage flux cutting the key bars. This flux creates a robust voltage that wants to drive current in the key bars.

Industry consensus points to current flow between adjacent key bars—that is, current flows down one key bar and returns through the adjacent key bar. However, the speaker proposed that the current flows as it does in an amortisseur winding—that is, down the key bars on one side of the stator and back 180 deg away on the opposite side.

In either case, the key-bar currents want to connect through three parallel paths: core flange, core laminations, and frame components. OEMs have tried to bridge these key-bar currents via shunting straps and silver plating (Fig 15). But regardless of shunting attempts, the voltage is robust and will want to flow current in all three paths. This leads to the local burning between key bars and core iron shown in Fig 16.

If this current flows only between adjacent key bars, it should not be of serious concern—that is, burn damage would be at the core OD where the laminations already are shorted via the key bars (on the common uninsulated key-bar structure).

However, a large generator of new design recently failed from core-iron melting at the bottom of the slot. Apparently, this failure was very similar to the Duke unit where failure was caused by the over-fluxing discussed earlier by Chairman Smith. Core failure at the bottom of the slots is exactly where failure would be expected if the key bars act as an amortisseur winding.

There are eight more units similar to this failed generator in the US and others have the key-bar-to-core-iron heating (Fig 17). Maughan expressed hope that this all would be better understood soon. Access presentation.

Testing of wedge tightness and stator cores

Mladen Sasic, IRIS Power

Sasic’s presentation included well-received tutorials on both wedging and core function. Integrity evaluation of both components also was included. With respect to stator slot wedging, typical tests include visual inspection for wedge or filler migration, evidence of greasing or dusting, ripple-spring compression (on some designs), and wedge-tightness assessment by manual or mechanized tapping.

Wedge tightness is a subjective call: There is no industry-wide agreement on what constitutes a tight wedge. For a particular mechanized test device, a Relative Tightness Index can be assigned by the manufacturer, but only for that device. If it can be concluded that the output is meaningful, then this reading can be important both for assessing the tightness of individual wedges and for assessing the overall wedge-tightness spectrum between subsequent tests.

But for manual testing, the accuracy of results depends completely on the skill and experience of the individual performing the test. The bottom line: There is a lot of uncertainty with any method. Personal “feel” often is considered more accurate than instruments; there is no agreement on what constitutes tight and loose; and introduction of online methods may be helpful.

Relative to core-tightness testing, inspection is important in the search for evidence of local dusting or greasing. Suspected loose areas can be assessed using the so-called knife test: Will a knife enter between laminations with a recommended maximum force of 25 to 30 lb?

Relative to lamination insulation evaluation, two methods are common: high- and low-flux tests. On high-flux testing, a heavy excitation cable is used to pass the necessary high current to excite the core back iron to a level below rated flux. Eddy currents will flow in local spots where lamination insulation is degraded. The resulting heat is measured. But there is no agreement on excitation levels, test duration, and acceptance criteria.

There are several concerns with the high-flux test, including: dangers from use of high voltage and current, local core burning, general core overheating, temperature attenuation for damage deep in the core and costly in time, material, equipment, and labor.

The most commonly used low-flux test is ElCid. It has numerous advantages over high-flux testing, including: low power and voltage, low risk to personnel or core, fast and easy, hand portable equipment, instant interpretation of results, convenient to perform repetitively, can be done with rotor in place.

But there are disadvantages to low-flux testing, too. It requires competent, trained personnel; output can be misleading, and it has an imperfect correlation with the high-flux test. Access presentation.